wiki-stats/wiki_stats.py

import os
import re
import pandas as pd
from lxml import etree
import matplotlib.pyplot as plt
import mwxml  # pip install mwxml

OUTPUT_DIR = 'csv'
PROJECTS_TEAM = [  # Oderered by first entry
    'Anderson',
    'Robert',
    'Brett',
    'Conrad',
    'Matt',
    'Duncan',
    'Ianc',
    'William',
    'Laurent',
    'Alexandra',
    'Wendy',
    'Alessio',
    'Luis',
    'Grantley',
    'Beatrice',
    'Sarah',
    'Rossm',
    'ROBERT',
    'Steven',
    'Wendyth',
    'Melissa',
    'Andy',
    'Michael',
    'Msa',
    'Jamie',
    'Toms',
    'Sam',
    'Larry',
    'Annakg',
    'Hamish',
    'Francois',
    'Annab',
    'Erica',
    'Coral',
    'Priom',
    'Barry',
    'Nathan',
    'Chrisd',
    'Andrewf',
    'Joshuas',
    'Daniel',
    'Danh',
    'Duncanr',
    'Robertt',
    'Chrisdu',
    'Brettm',
    'Mathieud',
    'Ianco',
    'Larryp',
    'Grantleys',
    'Aliceh',
    'Mattb',
    'Tobyt',
    'Benm',
    'Jamess',
    'Douga',
    'Gabil',
    'Francoisf',
    'Lluna',
]


def main(xml_name):
    revisions = []
    with open(xml_name, 'rb') as f:
        pages = mwxml.Dump.from_file(f)
        for page in pages:
            title = page.title
            for revision in page:
                timestamp = str(revision.timestamp)
                contributor = revision.user.text

                revisions.append({
                    'page': title,
                    'user': contributor,
                    'date': timestamp
                })

    df = pd.DataFrame(revisions)
    df.index = pd.to_datetime(df['date'])

    # Convert to local time, and round to nearest day
    df = df.tz_convert('Australia/Sydney')
    df = df.sort_index()
    df['date'] = df.index.date

    # Drop duplicate entries (ignore edits for same user/same page/same day)
    df = df.drop_duplicates()
    df = df.drop(columns=['date'])

    # Only include non-academic users
    df = df[df['user'].isin(PROJECTS_TEAM)]

    # Get list of years
    years = df.index.year.unique()

    os.makedirs(OUTPUT_DIR, exist_ok=True)

    summary = []
    for year in years:
        idx = df.index.year == year
        pages = df[idx].groupby('page').count()
        pages = pages.rename(columns={'user': 'edits'})
        pages = pages.sort_values('edits', ascending=False)

        users = df[idx].groupby('user').count()
        users = users.rename(columns={'page': 'edits'})
        users = users.sort_values('edits', ascending=False)

        pages.to_csv(f'{OUTPUT_DIR}/{year}-pages.csv')
        users.to_csv(f'{OUTPUT_DIR}/{year}-users.csv')

        summary.append({
            'year': year,
            'page edits': pages.shape[0],
            'active users': users.shape[0]
        })

    summary = pd.DataFrame(summary)
    summary = summary.set_index('year')

    fig, ax = plt.subplots(2,
                           1,
                           figsize=(6, 4),
                           sharex=True,
                           gridspec_kw={'hspace': 0.5})

    summary[['page edits']].plot.bar(ax=ax[0], legend=False)
    summary[['active users']].plot.bar(ax=ax[1], legend=False)

    j = 0
    for i, row in summary.iterrows():
        ax[0].annotate(row['page edits'],
                       xy=(j, row['page edits']),
                       xytext=(0, 6),
                       textcoords='offset pixels',
                       ha='center',
                       fontsize=8)

        ax[1].annotate(row['active users'],
                       xy=(j, row['active users']),
                       xytext=(0, 6),
                       textcoords='offset pixels',
                       ha='center',
                       fontsize=8)
        j += 1

    ax[0].set_title('Page edits', fontsize=10, y=0.9)
    ax[1].set_title('Active users', fontsize=10, y=0.9)

    ax[1].set_xlabel('')
    ax[0].set_ylabel('Count', labelpad=10)
    ax[1].set_ylabel('Count', labelpad=10)

    for a in ax.ravel():
        a.spines['top'].set_visible(False)
        a.spines['right'].set_visible(False)
        png_name = xml_name.replace('.xml', '.png')

    plt.savefig(png_name, bbox_inches='tight', dpi=300)


if __name__ == '__main__':
    xml_names = [f for f in os.listdir('.') if f.endswith('.xml')]
    for xml_name in xml_names:
        main(xml_name)